Special considerations for drone surveying in mines
In addition to determining which drone solution would be best to use, Cazorla also identified some key considerations specific to mining that impacted the flight plan.
Because the terrain for this project changes significantly at times – the elevation is approximately 150 m – Cazorla used a function on the drone called terrain follow, which ensures that the drone stays at a constant altitude during the flight. Under different circumstances, when surveying a natural mountain, terrain follow can be used in default mode. However, this functionality must be approached differently in a mine.
“Mines are a little bit special,” said Cazorla. “Because the elevations we are surveying are artificially made, we cannot just use the terrain following function in default mode. Areas with frequent movement of materials over short periods of time, like mines and construction sites, require an up-to-date DSM be inputted into mdCockpit, the Micro-drones flight planning software, to use the terrain following option. By default, mdCockpit uses a global terrain model SRTM, which may not represent the current reality of a constantly-changing terrain.”
Once that DSM is inputted and terrain following is selected, the us-er can then choose the proper altitude to fly – another detail with special considerations for mining. In choosing the height, the user must consider the reflectivity of the surface, which varies depending on the rocks and minerals present. “Coal has a similar reflectivity to asphalt, about 17%,” explained Cazorla. “So, we needed to be a bit conservative in our choice of altitude. For this project, we chose 50 m because that is the maximum altitude that could produce the required point cloud density with that expected surface reflectivity.”
According to Cazorla, a conservative overlap between flight strips is also desirable for mining projects, due to the irregular terrain and possible vertical walls. For this project, Cazorla set the overlap at 40% to ensure all necessary data was collected in high-slope areas.
Because this project was utilising LiDAR, no ground control points (GCP) were necessary. However, Cazorla incorporated a few targets, including one Smart Target with GNSS receiver, into the setup in order to help check the accuracy of the project for quality control during the processing phase.
Having set all these parameters, the project team was now ready to outline the timeframe for data collection. The team used the company’s mdFlight Performance App (mdFPA) to determine the maximum flight time per battery under their specific environmental circumstances and flight parameters. The total estimated flight time was 2 hours, including four flights of approximately 30 mins. To simplify the task, they made use of the Resume Waypoint feature in the mdCockpit app, which allows operators to change the battery mid-way through the flight and then have the flight automatically resume exactly where it left off.
The last part of mission planning is to calculate the expected project errors based on the system specifications. The expected accuracy of this project was approximately 6 cm. Later, during the processing phase, these calculations were used to check the accuracy of the out-comes.
All flight parameters for this project are summarised in Table 2.
|Table 2. Summary of flight parameters|
|Areas||1000 x 900 m|
|Path||Terrain following with 'up-to-date' DSM|
|Flight direction||Parallel to the slope|
|GCSs/Check points||2 targets|
|Time frame data collection expected||2 hours|
|Mission plan||Resume wypoint|
Data collection and quality control
“From there, we are ready to fly!” exclaimed Cazorla.
Once the drone launches, it autonomously follows the planned flight path, collecting data in strips while travelling back and forth over the area selected in the company’s application. The pilot monitors the flight by watching the drone both physically in the sky and virtually on a tablet.
In less than 2 hours, the data acquisition was complete and Cazorla was ready to begin processing. He implemented several quality control steps to ensure the accuracy of the data. First, he examined the GCPs, checking that they were all in the area of interest, that no target was partially covered or damaged and that there were no initial problems with any target. Once satisfied with the condition of GCP imagery, he processed the trajectory using mdInfinity, a processing software developed by the company.
“The first thing I look at is the processing status,” said Cazorla. “The closer to zero, the better it is. Then I check the RMS for the trajectory. In this case, it was less than 2 cm for X and Y and less than 3 cm for Z. Last, I check that there are no IMU gaps for any of the flights.”
Figure 1. Final mapping results: point cloud.
Having gone through these quality control steps, he then generated the georeferenced point cloud by inputting the raw LiDAR data and the payload calibrations into mdInfinity (Figure 1). Once the point cloud was generated, he used the precision enhancement tool to improve the point cloud, explaining that this step would improve the overall quality of the DSM.
From there, he inputted the point cloud into GIS software, selected the desired accuracy and elevation grid and created the DSM (Figure 2). He also created a contour line generation, another requirement of the coal mining company.
Figure 2. Final mapping results: digital surface model (DSM).
“Once the deliverables were created, we conducted a strip alignment analysis, an accuracy assessment and error analysis to compare what we got to the project requirements,” said Cazorla. “We found that the accuracy was very good and that the data and deliverables met all of the requirements. Our RSM was 1.5 cm, which is extremely accurate, even more than I would expect. Users of this system can expect an RSM less than 6 cm.”
Next, Cazorla created an independent accuracy assessment report of the 3D point cloud to confirm the quality of the project for the coal mining company and establish that the mission was successful.
“The process may sound complex, but once you have done it a few times, it is very easy,” concluded Cazorla. “Some people can be hesitant to learn new technologies, but I say it is worth it. You can learn to do all of this in the amount of time you would save using a drone on just one project.”
A more efficient workflow
Previously, to calculate soil volume at the coal mine, a two man surveying crew using GNSS RTK receivers would complete the project in 3 – 5 days. Deploying the Microdrones workflow and mdLiDAR1000 meant they were able to fly the entire project in approximately 2 hours and process the data the same day. Cazorla was therefore able to successfully demonstrate that drone-based LiDAR is much more efficient than traditional surveying methods.
The first part of this article is available here.
Read the article online at: https://www.globalminingreview.com/special-reports/22092020/go-with-the-workflow/
You might also like
Flender has received the Supplier Excellence Recognition Award from Caterpillar.